The influence of alcohol and related medical factors on an individual’s cognitive state poses significant risksin contexts such as driving, machine operation, and decisionmaking. Conventional alcohol detection systems rely mainly onBlood Alcohol Concentration (BAC) levels, which fail to capture variations in individual tolerance and broader physiologicalimpacts. To address this limitation, we propose an AIoT-Based Mental Fitness Detection System that integrates multiple lowcost biosensors, including the MQ-3 (alcohol), MAX30102 (heart rate and SpO2), and Galvanic Skin Response (GSR) sensor,interfaced with an ESP32 microcontroller for real-time data acquisition and wireless transmission. The collected multimodalphysiological signals are processed using a lightweight Machine Learning model (shallow neural network/SVM/Random Forest)to classify an individual’s state as Fit or Unfit. Unlike traditional alcohol detection devices, our system considers combinedfactors such as stress, anxiety, and oxygen saturation levels, offering a more holistic evaluation of mental fitness. A Flaskbased backend handles data processing and prediction, while a web dashboard provides intuitive visualizations of live sensorreadings, historical trends, and system decisions. Experimental validation demonstrates the feasibility of deploying this systemas a low-cost, portable, and scalable solution for real-time mental fitness assessment. This work contributes to enhancing road and workplace safety by shifting from single-metric alcohol detection to AI-driven multimodal fitness evaluation.
RAHUL et al. (Fri,) studied this question.